Possible modifications of the basic experimental design of the National Toxicology Program's two-year carcinogenesis bioassay are being investigated. Changes being considered include additional dose levels, reallocation of animals among experimental and control groups, and the selection of dose levels other than the traditional MTD (maximum tolerated dose) and 1/2 MTD. The purpose of these modifications would be to permit more precise risk estimates at low-dose levels without adversely affecting the power of the bioassay for detecting carcinogenic effects. Additional research efforts include the problem of incorporating historical control information into the statistical analysis of bioassay data.